Human Transit

how do you compare to your peers? should you care?

Admit it:  You've always cared, at least in secret, about how you compare to your peers: your friends, your fellow students, your graduating class, your co-workers, your generation.  Well, deep down, transit authorities and city governments care too, which is why comparing a city to other similar cities always gets attention.

Sometimes peer comparisons cause complacency, especially if you choose the wrong peers.  Wellington has the highest transit mode share in New Zealand, but in a country with only one other big, dense city, that obviously shouldn't imply that it's reached nirvana.  Working in greater Vancouver I always have to emphasize that they are doing so well by North American standards that they have to start comparing themselves to European port cities in their size class (Glasgow, Edinburgh, Copenhagen, Helsinki, Marseilles).  My general advice: If your peer comparison says you're wonderful, throw a party and revel in this for 48 hours, then look for a more motivating group of peers. 

At the other extreme, nothing is more motivating than being told that you're dead last among your peers.  Earlier this year I worked (through my Australian employer MRCagney under the leadership of Ian Wallis Associates) on a peer comparison study for Auckland, New Zealand, which compared Auckland's transit performance with all the five biggest Australian cities plus a selection of North American ones.  Download the full report here.  Remember, if you're in any of the peer cities that it uses (Wellington, Brisbane, Perth, Adelaide, Sydney, Melbourne, Edmonton, Ottawa, Calgary, Vancouver, Honolulu, Portland, Seattle) this is your peer study too!  Just keep the tables and refocus the text (citing the source of course!).

More generally, the report is a good illustration of how peer comparison can work at its best, and also of the cautions that must be shouted from the sidelines once the conclusions take fire in the media, as they certainly have in Auckland.  From yesterday's New Zealand Herald:

Consultants have ranked Auckland last out of 14 cities - in New Zealand, Australia, Canada and the United States - included in a benchmark study for the average number of public transport trips taken annually by its residents.

Aucklanders also pay the highest fares of any of the cities, amounting to 24c for every kilometre travelled on the average 44 public transport trips they take each year, compared with 17c in Wellington.

The rest of the article is further grim statistics, plus quotations from political leaders demanding that something be done. 

I'm  sympathetic to Auckland Transport in this case.  Remember, a city's transit performance is mostly about the physical layout of the city and the constraints on other modes; the quality of the transit system by itself can't overcome problems in those areas.  The nature of the economy also matters.  Wellington is much smaller but it has much more severe chokepoints in its urban structure.  In fact, all travel between the northern and southern parts of the city must go through a single chokepoint less than 1 km wide, which is also the (very dense) downtown.  Wellington's economy is dominated by government, which is generally a sector disposed to use transit heavily. All of these features are hugely important in driving Wellington's mode share above Auckland's, and yet they don't include anything about the respective quality of the transit systems. 

Peer comparisons also carry the false assumption that everyone wants to be the same kind of city, and is therefore working to the same kind of goals.  (This attitude, taken to extreme, produces the absurdity of top ten "best cities for transit" lists.)  Low mode share for transit may mean your transit system is failing, but it may mean that it's not trying for mode share, or at least that it has other objectives or constraints that prevent it from focusing on that goal.  It may just mean that your city has different values.  It may mean the city stikes a different balance between cycling, transit, and walking based on its own geography.

Still, service quality matters, and there's a lot that Auckland can do.  I hope the city's opinion leaders are listening to Auckland Transport as well as berating it, so that they understand the real choices that must be made to move Auckland forward.  If there's a real conversation, great things can be accomplished. 

updated: another weird way to measure "best cities for transit"

UPDATE:  This post in its original form happened when I had the Atlantic article article on the new Brookings report, but couldn't find the report itself.  As it turned out, the Brookings report is much smarter than the Atlantic article made it sound.  In particular, it appears to have been the Atlantic, not Brookings, who decided that this is a "10 best" story.  I've made minor corrections to revise the attribution, but watch soon for a series of posts responding to the Brookings report more fairly.  I'll have several things to say about it, some of them critical, over the next few days.    

The esteemed Brookings Institution US magazine The Atlantic says that America's best city for transit is ... <drumroll> ... Honolulu! 

Sooner or later, everyone ends up on a "10 best" list.  This time around (apart from Honolulu) Brookings seems to be reaching out to southwestern cities that feature vast grids of fast car-centered arterials spreading out across any available flat land.  Are those the best cities for transit?  Well, maybe it's just their turn.

The Brookings top 10:

1. Honolulu, HI
2. San Jose - Silicon Valley, CA
3. Salt Lake City, UT
4. Tucson, AZ
5. Fresno, CA
6. Denver, CO
7. Albuquerque, NM
8. Las Vegas, NV
9. Provo-Orem, UT
10. Modesto, CA

As regular readers of this blog understand, "best city for transit" is a meaningless term.  There are many ways to define "best," and we don't always agree with ourselves on which one matters, let alone agree with anyone else.

But the Brookings definition, as described in the Atlantic article, is especially perverse, even by "10 best" standards:

Brookings graded each city according to two criteria -- coverage (the share of Americans within 3/4 miles from a transit stop) and job access (the share of city jobs accessible within 90 minutes of transit) -- to determine the ten best performing cities for public transportation.

I have no idea what "within 90 minutes of transit" means, and the original report doesn't appear to be on the Brookings website.  [UPDATE:  Now I do.  Look for a post on this soon.]

But residential coverage, as a primary indicator of transit quality, is a very loaded way of thinking about "best," especially if you care about transit sustainability outcomes that depend on ridership.  Many people are within walking distance of a bus stop but not within walking distance of service that's remotely attractive in terms of frequency, speed and reliability.  (Even more are within "air distance," which is what Brookings seems to refer to.)

Due to this definition, the Brookings Atlantic list comes to focus on rather low-density, car-dependent cities that happen to have good transit coverage.  The cities listed have transit systems that are complete in terms of getting close to almost every home, but in low-density cities this is usually achieved by sacrificing frequency, speed, and even directness.  For example, if you really want to maximise your residential coverage without spending money, and thus satisfy the Brookings criterion, just design routes like this:

This is what "residential coverage" standards encourage transit planners to do!  So Brookings needs to explain how an abundance of transit routes that look like this could indicate a "best city for transit."

Then there's a pedestrian environment.  Many of these cities are hard to walk in, and transit especially often delivers you to a busy arterial where you'll have difficulty walking to nearby destinations.

Finally, of course, "residential coverage" is about how many people have access to the system, which has nothing to do with who finds it useful.  In fact, a transit system that's trying to maximize its relevance is always trying to push residential coverage standards down so that they can focus more service on dense areas where residents are more likely to want to build their lives around transit and other sustainable modes. 

UPDATE:  Having now perused the Brookings report, I stand by these last three paragraphs, but will expand on them soon.  I'm relieved to know that Brookings's framing of the issue is smarter than the Atlantic made it sound. 

Still, a report like this from a distinguished institution raises great opportunities to question some of the more common assumptions about how transit should be compared across cities.  So I take issue with their reliance on city limits and Metropolitan Statistical Areas, among other things, not to criticize Brookings but simply to encourage more nuanced and coherent explanations of this kind of statistical work.  More posts on these critiques in the future, starting here.

request for information: canadian ridership (updated)

Some colleagues and I have lately had occasion to compare public transit performance in young metro areas of 1-3 million, across North America and Australasia, and we've been turning up draft charts like this one (click to enlarge):

The x-axis is population, the y-axis is public transit trips per capita, per annum.  The red triangles are the Australian cities: Adelaide, Perth, Brisbane, Melbourne, and Sydney.  The purple circles are US cities: Honolulu, Portland, and Seattle.  The blue dots are Wellington and Auckland in New Zealand. 

But we're especially struck by the green dots, which represent Canada.  The four dots are Ottawa, Edmonton, Calgary and Vancouver, and all are at the top of this heap despite relatively low populations.  Even tiny Edmonton and Calgary are outperforming all the other cities studied, including Sydney and Melbourne. 

I've been asked to write a few paragraphs about why Canada's cities do so much better than very similar cities in the US or Australia/NZ.  I have my own speculations, but anyone with references or insights on this, please comment! 

UPDATE:  The consultant in charge of this project has reviewed the comments so far (Oct 8 0900 Sydney time) and observes:

 

Some rapid comments, having skimmed through the blog responses:

• Area boundaries. Always a problem. We have tried to cover the whole metro service areas in each case.  For SYD, this has meant including Newcastle and Wollongong etc, given the CityRail services extend to these areas. Similarly we have included Gold Coast and Sunshine Coast in the Brisbane figures.
  
  
• ** Linked v unlinked v return trips.  I doubt if this is a problem—we have aimed for unlinked trips (boardings) in all cases.
• ** Densities may well be one of the explanators of the differences, and we have estimates of these for the  North American metro areas (but not the Aus/NZ areas. I am reluctant to pursue these in more detail—as you well know, this is a minefield.

The whole topic would be good for a PhD—why has no done it already??

Good question!  More on this topic here!

ridership down in america? look deeper

It seems to be, and I'm sure the New Republic's Robert Puentes is right about the causes -- (1) recession-driven unemployment (which both reduces commute demand and reduces discretionary income) plus (2) the epidemic of service cuts, which is proving yet again that not many riders are so "captive" that you can't drive them away eventually.  Both of those factors are well-observed correlations.

But it's interesting that the New Republic chose to feature this map from the Brookings Institution's interactive source, which shows total numbers of public transit commuters by metro area, as opposed to this one, which shows the percentage of all commutes that go by public transit, or what's technically called the Journey-to-Work (JTW) mode share.

A map of net numbers of transit commuters is, of course, mostly a map of how big cities are, which is just not very interesting.  

But if you look closely at the map of mode shares (which I can link to but can't seem to copy) you see a map of how dense big cities are.  Toggle the two, and as you go from raw commute numbers (the map above) to mode shares, you see the metro areas of Los Angeles, Houston, Atlanta, and Miami all shrink, which means that those areas have low public transit use compared to their big populations.  What doesn't change is the high score of the big, denser metros: the Northeast Corridor, Chicago, Seattle, and the San Francisco Bay Area.  Of course, those metros all have lots of sprawl, but their transit numbers are driven by very, very high ridership in their dense core cities, plus the heavy commute market into those cores due to the difficulty of driving and parking there.

All of which points to a couple of typical warnings about American transit statistics:

• Metro areas, by definition, conflate a dense core city and a lot of low-density suburbs, and thus conceal the dramatic difference between transit's relevance in those places.  Metro area statistics may be useful for judging the relative economic performance of different parts of a country, but they always conceal most of what matters in understanding how transit really works.  If you want to understand how and why transit works and fails, you should spend less time comparing metro Los Angeles and metro New York and more time comparing Manhattan with Nassau County, or comparing the LA/Hollywood core with the western San Fernando Valley.
• Even city limits of core cities are not always helpful.  City limits are a decent approximation of the core city in most of the northern tier of states, but in Southern California and Texas, for example, core cities are bizarre shapes that omit some dense inner city fabric and incorporate a lot of suburbia and exurbia.  I have yet to meet anyone who's personal conception of 'Los Angeles' ends at the bizarre Los Angeles city limits (thus including the San Fernando Valley and San Pedro, for example, but excluding West Hollywood, Santa Monica, and East LA).  The exception, of course, would be City of Los Angeles leaders and employees.  I've met many people in New York City, Seattle and San Francisco who's mental notion of their city matches the city limits pretty exactly.

Back to the Brookings maps, it's also worth noting that as you toggle, a lot of interesting smaller cities jump out -- cities with high rates of transit use for their relatively small populations:  Salt Lake City, Albuquerque, Tucson, Honolulu, New Orleans, Madison, and many others.  A mixed bag, but why those cities? 

These cities may remind you of the factors discussed here regarding low-car cities, where the dominant forces seemed to be city age, poverty, and dominant universities.  But when it comes to high mode-share small cities the factors are a little different. 

Poverty doesn't seem to matter at much on this score.  Poverty keeps people from owning cars but poor people also have shorter commutes, if they are fortunate enough to commute at all.  Poverty is generally aligned with low skills, which aligns with jobs that are standardized and interchangeable (e.g. McDonalds) as opposed to specialized (e.g. bioengineering), and the one great thing about a standardized and interchangeable job is that if you can find it anywhere, you can probably find it close to home.  Long-distance commutes are a problem of more educated workers, whose jobs are more specific and thus harder to find .  

But dominant universities, I'm willing to bet, matter a lot.  You may not think of Tucson as an especially transit-oriented place, and if you're not associated with the University of Arizona there, then it probably isn't for you.  But a massive university is the best transit demand engine a city can have: huge masses of temporarily poor people, often with strong environmentalist values, who come and go at the campus all day, not just in expensive-to-serve peak commute periods.

The second best market after a university is government employment, as governments are more motivated to help employees make good commute choices than private employers are.  So state capitals (Salt Lake, Honolulu, Madison) may surge a bit on that score.

Density, though, always matters, even in relatively small but mostly compact cities like Honolulu. 

Transit infrastructure investment matters too, but overall quantity and quality of service may matter more than the presence or absence of rail transit, for example.  Rail-heavy Salt Lake City doesn't stand out in comparison to similar but rail-less Tucson, for example.  That's not an argument against light rail.  Salt Lake's light rail line to the University of Utah, for example, is a great way of handling the huge capacity demands there.  But it's a warning that big transit infrastructure still responds to demand rather than creating it, especially in the short-term scale of the Brookings analysis. 

honolulu: grand themes from the rail transit wars

 

Eight months ago, a freelance reporter asked for my views on the emerging argument over Honolulu's proposed rail transit line, which would stretch most of the length of the populated southern shore, from west of Pearl Harbor through downtown to Ala Moana Center on the edge of Waikiki.  The Transport Politic has covered the background here and here and here.  A good blog on the subject is here.

A mostly-elevated driverless metro, quite similar to Vancouver's SkyTrain, is the preferred alternative.  Unlike SkyTrain, though, it would not go underground downtown.  Instead, a substantial elevated structure is proposed, one that would definitely have some visual impact, as this computer generated image indicates.  (For more attractive images of a proposed station, see here.)

Late in the process, long after this option had been selected and the environmental work done, there was a revolt.  One key player in the revolt was the local chapter of the American Institute of Architects, who have devoted a section of their website to attacking the project.  They propose a light rail solution where parts of the line could run at grade, and which would go through downtown Honolulu on-street at low speeds with frequent stops, much like the downtown light rail segments in Portland, San Diego, Denver, and many other cities.  Last year, a study by transit consultant Philip Craig came out arguing for this option.

If you care about the eternal struggle for planning power between the architecture and transport professions, you'll love this dispute.  It's perfectly designed to heighten the contrast between transportation thinking and urban-design thinking, and to confirm the worst stereotypes each holds of the other.  The transportation solution -- a mostly elevated driverless metro -- is fast, frequent, and reliable, but it does require a pretty impactful elevated structure (presumably because the best solution, undergrounding it through downtown, is too expensive).  The urban design solution, surface light rail, is beautiful downtown -- redolent of Portland, in fact -- but unfortuately it's also slow, less frequent, and less reliable.  

The most striking thing about the debate, for me, is that it's been discussed on both sides as though they main payoff of the elevated line is speed, rather than frequency, an example of the "travel time" fallacy I discussed here. 

Speed is certainly a big difference: end-to-end travel times on the surface light line are about 20-30% longer than on the elevated line, and any modeller will tell you that's enough to make a difference in ridership.  Moreover, the low speed is concentrated in the slow downtown segment, so medium-length trips that have to go through that segment could be more than 30% longer.   

But the frequency difference is even more profound.  Driverless metros are easy to run frequently all the time, even late in the evening, while surface light rail, which requires drivers, will always be under financial pressure to run the minimum possible amount of service at low-demand times.  I guarantee that if the surface light rail is built, it will never run better than every 15 minutes at 11:00 PM, but the driverless metro is easy to run every 5-6 minutes even when demand is low, because driverless metros sever the link between frequency and operating cost.

Reliability will also unquestionably be superior on the elevated line, because it will have no interference from other traffic.  Surface light rail, even if separated from traffic, still has to interact with it at intersections, and these cause unpredictable delay as well as a much higher rate of accidents.  Browsing Mr. Craig's report, though, I noticed alarming suggestions that the line could mix with car traffic at certain points.  It even suggests operating in the curb lane of some streets, exposing the line to constant interference with right-turning traffic, and all the accident opportunities that this will present. 

(At one point, the Portland Streetcar's two-year construction time is cited as a reason light rail can be fast.  This should be a screaming red flag, because it implies a common and dangerous categorical confusion between streetcar and light rail.  The Portland Streetcar is a fine example of a local-stop, low-speed circulator, but it is, by design, slow and unreliable.  It was also built on the cheap with more opportunity for disruption:  Proper surface light rail construction requires relocating utilities from under the line so that the line doesn't have to be shut down to access them later.  The Portland Streetcar was built by digging barely two feet below the pavement, leaving all the utilities in place.)

Finally, the AIA suggestion of putting light rail through downtown's current "transit mall," Hotel Street, seems to assume that buses somehow just get out of the way.  There's no thought on what the light rail does to the bus system, how the bus system would be revised to work with it, and what impacts this would have.  This is an example of an almost universal fallacy in thinking about rail transit outside the transit profession: that somehow, rail just makes buses disappear. 

The truth is more complex and interesting.  Rail replaces some buses but never all of them.  Major bus corridors still need to run into the city, and need to be accommodated there.  Even with light rail on the Portland transit mall, for example, buses are by far the majority of vehicles there, and the great design challenge of that street was how to integrate the light rail and bus services without letting them get in each other's way.  It's not an easy problem.  There are many solutions, but all of them have enough impacts that you have to think about this when you're proposing your rail line, not just assume that bus people will figure it out later.

None of this is an endorsement of the current elevated plan.  I have questions about it, too.  One of the few transportation advantages of the surface light rail option is that it's easy to imagine extending it into Waikiki, Honolulu's premier tourism district, in the future.  (The line as now proposed ends at Ala Moana Center, within walking distance of Waikiki but not quite there.)  I have a lot more trouble imagining extending an elevated line into such a tourist-intensive and view-sensitive area.

Finally, as a planner, I have to share the exasperation of everyone at the City of Honolulu who brought this project through all its necessary stages, including extensive public consultation, only to encounter this objection at the eleventh hour.  The late-breaking nature of the objection is a major strike against it, because it implies substantial delay to a project that was close to moving ahead.  Claims that the surface light rail will be cheaper than the elevated line may well turn out to be true, but you don't really know this until you've explored all the consequences of your project (such as where to put the buses!) and added the costs for all those effects and mitigations.  That's what the environmental planning process is for, and only the elevated line has been through that process.

Before you comment, please note:  My purpose here is to note some of the interesting large themes raised by this dispute, not to wade into the dispute itself.  I am quite intentionally expressing no view on the politics surrounding the line, or on the funding challenges.  I do not have a view on which option is best.  In fact, the tradeoff between efficiency and aesthetics is one of those value-judgment decisions that don't have a technical answer.  In a sense, nobody outside of Honolulu should have an opinion, because it's all about what kind of city they want, and that has to be up to them.

frequency and freedom on driverless rapid transit

If you've seen much of Vancouver on television the last few days, you've probably seen a shot of a small train gliding along an elevated guideway.  It's SkyTrain, the world's largest system of fully automated (driverless) metros.  Perhaps you've ridden driverless trains that shuttle between airport terminals.  SkyTrain is the same principle, at a citywide scale. 

Driverless trains raise all kinds of anxieties. Many people like knowing there's someone in charge on the vehicle, and imagine that this person will be useful in emergencies.  But on most subways, you can only talk to this person by pushing an intercom button, and there's very little he can do if there's an emergency in your car other than call for help. 

SkyTrain provides exactly the same emergency intercom button, but the person you're talking to is in a central control center instead of on the train.  As a result, they are likely to have a better view of the whole emergency, and the resources at hand, than a train driver could have.  There are still staff roaming the system, checking fares and providing assistance, but not one on every train.

Of course, if zoom and whoosh matter to you, there's nothing like the experience of riding in the single seat at the front of the train, enjoying a view that's usually reserved for a train driver.  And it's quite a view.

None of these, however, are the real reason to consider driverless metros.  The real reason is that they break the connection between frequency and labor costs.  Look at SkyTrain's frequency tables:

On the Expo and Millennium lines, you'll never wait more than 8 minutes for a train.  On the combined portions of these lines, which connect some of the densest centers in greater Vancouver, you'll never wait more than 4 minutes -- not even at 7:00 AM on Sunday or 11:45 PM on a Tuesday.    In a year of living in Vancouver and riding Skytrain every day, I never waited more than 5 minutes. 

Some non-automated subways do run as frequently as SkyTrain, but they're in places like Manhattan, London, and Paris, cities vastly larger than greater Vancouver.  (Paris, by the way, now has a driverless line, and is beginning to convert existing lines to driverless.)  No non-automated system could have delivered such high frequencies late into the night in a city of Vancouver's scale. 

The lack of a driver is the key to those extreme frequencies.  When you have a driver on every vehicle, the labor cost is the dominant cost of operations.  So when you have to cut service, as many North American agencies are doing this year, you end up cutting frequencies, starting with late night and weekend.  Many North American light rail systems are dropping below even a 15-minute frequency in the evening, making themselves increasingly useless for the spontaneous trips that are essential to freedom in urban life.

As I've followed the debates between driverless metro and light rail, most recently in Honolulu, I've been struck by how rarely frequency is mentioned.  Reading the coverage of Honolulu's debate between elevated driverless metro and light rail, you'd think that the primary question is profile -- elevated vs surface -- and that this is a question of running fast on an ugly elevated structure or running slowly but more accessibly on street.  Few on either side make the crucial point that a driverless system will be able to run frequently even very late at night, early in the morning, whenever you want to travel.  Light rail, bound by labor costs, will always be pressured to cut frequency outside the peak period. 

This is not, in any sense, a blanket endorsement of driverless metros over light rail.  They are different tools for different situations.  The stable operating cost of driverless metros comes at a capital cost.  Vancouver's next project, the Evergreen Line, is expected to cost $1.4 billion (CAD) for 11 km, well over twice the per-km cost of typical light rail. 

Light rail is wonderfully flexible, able to run onstreet with signalized intersections, and across pedestrian zones, as well as in conventional elevated or underground  profiles.  Driverless metro must be totally grade-separated, which in practice usually means elevated or underground.  SkyTrain got its name because the original lines were mostly elevated, though the newest, the Canada Line, has a long underground segment. (I'll do another post on the local community impacts of the SkyTrain elevated structure, and the serious problems (and opportunites) they create.  I'll do yet another on Vancouver's current SkyTrain vs light rail debate, in its Broadway corridor.)

As you'd expect, SkyTrain has higher average speeds and reliability than surface light rail.  As always in urban transit, speed and reliability are mostly not features of the technology; they are mainly about how much you stop and what can get in your way, and not much can get in the way of a fully grade-separated system like SkyTrain.    

But I am struck that in these debates there's so little talk of off-peak frequency.  As an urbanist, I want to see vibrant cities that are full of activity for 18 if not 24 hours a day.  That means I want nightlife.  Isn't that a good reason to be interested in a service that really can run every 4 minutes at midnight? 

Last Photo:  Gordon Price, Price Tags

is elevated acceptable?

The Transport Politic has an excellent post on the debate over the plan to build Honolulu's proposed light rail system elevated through downtown, as opposed to at the surface as a group of architects wants. 

Everyone is prone to reduce the complexity of urbanism to a problem solvable by their own profession, and risks being dismissive of the expertise of other professions' points of view.  (See here, for example.)  When a group of architects proposes that a major new transit investment should be made slower and more expensive to operate in order to foster a better streetscape, as is happening in Honolulu, one hopes that they have thought through the urbanist consequences of all the people who'll be in cars instead of on transit because the transit is too slow, infrequent, and unreliable.  Let me clarify each of those words: